mysql5.7的主从切换 主库： master 10.11.0.211 从库： slave 10.11.0.210 目标：主从切换，将slave切换为master，master切换为slave 一、使用innobackup进行主从同步设置 # 应用的连接数据库的账号配置 grant select,insert,update,delete on task.* to task_user@"%" identified by "cmslU6WFkX2pBylwINy2T"; # 修改root密码 update mysql.user set authentication_string = password('root'), password_expired = 'N', password_last_changed = now() where user = 'root'; # 设置主从同步： # 备份主库 innobackupex --defaults-file="/etc/my.cnf" --user=root -p'root' --socket=/tmp/mysql.sock --apply-log --use-memory=1G /opt/2019-12-06_15-51-53 # 应用事务 innobackupex --defaults-file="/etc/my.cnf" --user=root --socket=/tmp/mysql.sock --copy-back /data/2019-12-06_15-51-53/ # 配置主从账号 grant replication slave,reload,super on *.* to rep@'%' identified by 'wsdb123'; # 记录主库的同步位置 [root@node1 data]# cat 2019-12-06_15-51-53/xtrabackup_binlog_info mysql_bin.000003 154 # 设置主从同步 change master to master_host='10.11.0.211',master_user='rep',master_password='wsdb123',master_log_file='mysql_bin.000003',master_log_pos=154; 二、进行主从切换操作: 1.锁定主数据写操作： master: mysql> flush tables with read lock; 2然后修改从数据库为主要数据库： 1> 保证所有从数据库都已经执行了relay log中的全部更新，在从服务器中执行stop slave io_thread,用show processlist检查，查看状态是否是Has read all relay log,表示更新完成. slave: mysql>stop slave; mysql>stop slave io_thread; mysql>show processlist\G *************************** 2. row *************************** Id: 2 User: system user Host: db: NULL Command: Connect Time: 4757 State: Has read all relay log; waiting for the slave I/O thread to update it Info: NULL 2> 在从服务器上执行stop slave，reset master命令，重置成主数据库 mysql>stop slave; mysql>reset master; mysql>reset slave all; -- 清除同步信息 3> 删除新的主服务器数据库目录中的master.info和relay-log.info文件，否则下次重启时还会按照从服务器来启动， 关闭新主库的my.cnf配置 read_only 4> 在新的主库中查看binlog日志 mysql> show master status\G; mysql> show master status\G *************************** 1. row *************************** File: mysql_bin.000001 Position: 154 Binlog_Do_DB: Binlog_Ignore_DB: Executed_Gtid_Set: 重新设置主从同步， 附切换从数据库命令: change master to master_host='10.11.0.210',master_user='rep', master_password='wsdb123', master_log_file='mysql_bin.000001', master_log_pos=154; start slave; 5>从服务器上检测是否复制正常(Slave_IO_Running: Yes && Slave_SQL_Running: Yes) mysql> start slave; Query OK, 0 rows affected (0.03 sec) mysql> show slave status\G *************************** 1. row *************************** Slave_IO_State: Waiting for master to send event Master_Host: 10.11.0.210 Master_User: rep Master_Port: 3306 Connect_Retry: 60 Master_Log_File: mysql_bin.000001 Read_Master_Log_Pos: 154 Relay_Log_File: mysql-relay-bin.000002 Relay_Log_Pos: 320 Relay_Master_Log_File: mysql_bin.000001 解锁定原主数据写操作： mysql> unlock tables; Query OK, 0 rows affected (0.00 sec) #修改原主库的my.cnf 打开read_only参数 进行相应的应用检查 数据库的测试： 主库创建表，并插入数据： MySQL [wanxing]> use wanxing; Database changed MySQL [wanxing]> show tables; +-------------------+ | Tables_in_wanxing | +-------------------+ | users | +-------------------+ 1 row in set (0.00 sec) MySQL [wanxing]> create table itpart(id int,name varchar(30)) ; Query OK, 0 rows affected (0.01 sec) MySQL [wanxing]> insert into itpart(1,'it engineer'); 从库查询数据是否同步： mysql> select * from itpart; +------+-------------+ | id | name | +------+-------------+ | 1 | it engineer | +------+-------------+ 1 row in set (0.01 sec)
生产环境切换记录：
主库和新主库都开启screen 窗口进行操作，避免网络中断引起操作异常 screen -S changemaster 准备工作：停止数据库的写入程序 停止监控 1.锁定老的主库写操作： master: flush tables with read lock; set global read_only=on; show variables like 'read_only'; show master status\G; 2.然后修改从数据库为主要数据库： 1> 保证所有从数据库都已经执行了relay log中的全部更新，在从服务器中执行stop slave io_thread,用show processlist检查，查看状态是否是Has read all relay log,表示更新完成 slave: mysql>stop slave; mysql>stop slave io_thread; mysql>show processlist\G *************************** 2. row *************************** Id: 2 User: system user Host: db: NULL Command: Connect Time: 4757 State: Has read all relay log; waiting for the slave I/O thread to update it Info: NULL 在从服务器上执行stop slave，reset master命令，重置成主数据库 mysql>stop slave; mysql>reset master; mysql>reset slave all; # 打开可写操作 set global read_only=off; 3> 删除新的主服务器数据库目录中的master.info和relay-log.info文件，否则下次重启时还会按照从服务器来启动， 关闭新主库的my.cnf配置 read_only 注释掉 #### 修改/etc/hosts配置，进行 # db 172.16.0.239 inf.nei.prod.mysql.eus 4> 在新的主库中查看binlog日志 mysql> show master status\G *************************** 1. row *************************** File: mysql_bin.000001 Position: 154 Binlog_Do_DB: Binlog_Ignore_DB: Executed_Gtid_Set: 重新设置主从同步， 附切换从数据库命令: # 日志标记 change master to master_host='10.11.0.210',master_user='rep', master_password='wsdb123', master_log_file='mysql_bin.000001', master_log_pos=154; start slave; # 从库中执行 mysql> change master to master_host='172.16.0.239' ,master_user='rep',master_password='wsdb123',master_auto_position=1; Query OK, 0 rows affected, 2 warnings (0.30 sec) mysql> start slave; Query OK, 0 rows affected (0.02 sec) 5>从服务器上检测是否复制正常(Slave_IO_Running: Yes && Slave_SQL_Running: Yes) mysql> start slave; Query OK, 0 rows affected (0.03 sec) mysql> show slave status\G *************************** 1. row *************************** Slave_IO_State: Waiting for master to send event Master_Host: 10.11.0.210 Master_User: rep Master_Port: 3306 Connect_Retry: 60 Master_Log_File: mysql_bin.000001 Read_Master_Log_Pos: 154 Relay_Log_File: mysql-relay-bin.000002 Relay_Log_Pos: 320 Relay_Master_Log_File: mysql_bin.000001 解锁定原主数据写操作： mysql> unlock tables; Query OK, 0 rows affected (0.00 sec) 注意： 1.修改原主库的my.cnf 打开read_only参数，和relay-log相关参数 relay-log = mysql-relay-bin relay-log-index = relay.index 2.注意新主库的read_only参数是动态修改的，一定要修改my.cnf为off，或者直接删掉，避免重启后为on，应用不可写 进行相应的应用检查 后续工作： 1.数据库备份并且做主从同步 2.从库上做冷备份 添加监控 主库：8核32G配置参考 [:~]# cat /etc/my.cnf
# Example MySQL config file for medium systems.
# 8核 16G
# This is for a system with little memory (16G) where MySQL plays
[client]
#password = your_password
port = 3306
socket = /tmp/mysql.sock
# The MySQL server
[mysqld]
user = mysql
port = 3306
bind-address = 0.0.0.0
socket = /tmp/mysql.sock
datadir = /data/mysql_data
pid-file = /data/mysql_data/mysql.pid
basedir = /usr/local/mysql
tmpdir = /tmp
#此开关默认为NULL，即不允许导入导出。
#secure-file-priv = /opt/upload
#-------------------------------gobal variables------------------------#
#默认关闭，涉及到timestamp类型的列自动更新的问题
explicit_defaults_for_timestamp = 1
###transaction_write_set_extraction = XXHASH64 #以便在server收集写集合的同时将其记录到二进制日志。并且是行更改后的唯一标识此标识将用于检测冲突。
###loose-group_replication_group_name = 'ce9be252-2b71-11e6-b8f4-00212889f856' #组的名字可以随便起,但不能用主机的GTID
###loose-group_replication_start_on_boot = off #为了避免每次启动自动引导具有相同名称的第二个组,所以设置为OFF。
###loose-group_replication_bootstrap_group = off #同上
###loose-group_replication_local_address = '192.168.1.88:33071' #写自己主机所在IP
###loose-group_replication_group_seeds ='192.168.1.88:33071,192.168.1.89:33071,192.168.1.90:33071'
###loose-group_replication_single_primary_mode = off #关闭单主模式的参数
###loose-group_replication_enforce_update_everywhere_checks = on #开启多主模式的参数
skip-external-locking
skip-name-resolve
skip-ssl
#memory is 16G
key_buffer_size = 16M
table_open_cache = 2048
table_definition_cache = 1024
sort_buffer_size = 4M
net_buffer_length = 32K
read_buffer_size = 4M
read_rnd_buffer_size = 16M
open_files_limit = 10000
thread_cache_size = 400
query_cache_type = 0
query_cache_size = 32M
max_write_lock_count = 300
wait_timeout = 28800
interactive_timeout = 28800
net_read_timeout = 1200
net_write_timeout = 1200
max_connections = 800
max_user_connections = 750
max_connect_errors = 10000
max_allowed_packet = 256M
back_log = 2048
log_timestamps = system
performance_schema = OFF
character_set_server = utf8mb4
##当链接数耗尽后，通过设置别用端口，让root可以登录
extra_max_connections = 2
extra_port = 13306
###让mysql不区分大小写敏感
lower_case_table_names = 1
#explicit_defaults_for_timestamp = 1
#----------------Myisam--------------------------------#
myisam_recover_options = DEFAULT
bulk_insert_buffer_size = 32M
myisam_sort_buffer_size = 64M
myisam_max_sort_file_size = 256M
myisam_repair_threads = 1
#if the query is exec time great than 2 seconds, the query will log to slow log if slowlog is enabled.
long_query_time = 3
slow_query_log = On
slow-query-log-file = /data/mysql_data/slow.log
show_compatibility_56 = on
# Don't listen on a TCP/IP port at all. This can be a security enhancement,
# if all processes that need to connect to mysqld run on the same host.
# All interaction with mysqld must be made via Unix sockets or named pipes.
# Note that using this option without enabling named pipes on Windows
# (via the "enable-named-pipe" option) will render mysqld useless!
#
#skip-networking
#----------------------------MySQL Log----------------#
# Replication Master Server (default)
# binary logging is required for replication
log-bin = mysql-bin
expire_logs_days = 10
log_error = error.log
log_error_verbosity = 1
log_warnings = 1
# binary logging format - mixed recommended
binlog_format = row
relay-log = mysql-relay-bin
relay-log-index = relay.index
# required unique id between 1 and 2^32 - 1
server-id = 239
#sql-mode = "NO_AUTO_CREATE_USER,NO_ENGINE_SUBSTITUTION"
#sql-mode = "NO_AUTO_CREATE_USER,NO_ENGINE_SUBSTITUTION"
sync_binlog = 1
log_slave_updates = 1
#binlog_checksum = NONE
#slave_skip_errors = 1062,1032
#------------------------------replicate--------------#
#排除不需要同步的库表
#replicate-ignore-db = mysql
#replicate-ignore-db = sys
replicate-ignore-db = information_schema
replicate-ignore-db = performance_schema
replicate-ignore-db = undolog
replicate-ignore-db = for_nagios
replicate-ignore-db = undolog
#replicate_wild_ignore_table = mysql.%
#replicate_wild_ignore_table = sys.%
replicate_wild_ignore_table = information_schema.%
replicate_wild_ignore_table = performance_schema.%
replicate_wild_ignore_table = undolog.%
replicate_wild_ignore_table = for_nagios.%
replicate_wild_ignore_table = undolog.%
#主主复制需要开启
#auto_increment_offset= 2
#auto_increment_increment= 2
#GTID模式复制，需要开启如下
gtid_mode = ON
enforce_gtid_consistency = ON
#并发复制
slave-parallel-type = LOGICAL_CLOCK
slave-parallel-workers = 2
master_info_repository = TABLE
relay_log_info_repository = TABLE
relay_log_recovery = ON
#跳过slave进程启动参数
skip-slave-start
#如果实例为从库，则需要设置为on
#read_only = off
#skip-grant-tables
#--------------------------------------------------------innoDB------------#
innodb_rollback_on_timeout
# Uncomment the following if you are using InnoDB tables
innodb_data_home_dir = /data/mysql_data
innodb_data_file_path = ibdata1:1G;ibdata2:1G:autoextend
innodb_log_group_home_dir = /data/mysql_data
innodb_undo_directory = /data/mysql_data/undolog/
innodb_undo_logs = 128
innodb_undo_tablespaces = 3
# You can set .._buffer_pool_size up to 50 - 80 %
#innodb_use_sys_malloc = 0
#innodb_page_size = 8192
innodb_buffer_pool_size = 12G
innodb_buffer_pool_instances = 1
#innodb_additional_mem_pool_size = 8M
# Set .._log_file_size to 25 % of buffer pool size
innodb_log_file_size = 256M
innodb_log_buffer_size = 64M
innodb_log_files_in_group = 3
#每次事务提交时MySQL都会把log buffer的数据写入log file，并且flush(刷到磁盘)中去，该模式为系统默认
innodb_flush_log_at_trx_commit = 1
innodb_lock_wait_timeout = 120
#启用独立表空间
innodb_file_per_table = 1
#CPU是1颗8核的，那么可以设置
innodb_read_io_threads = 8
innodb_write_io_threads = 8
#默认是0，则表示没有并发线程数限制，所有请求都会直接请求线程执行，当并发用户线程数量小于64，建议设置innodb_thread_concurrency=0，在大多数情况下，最佳的值是小于并接近虚拟CPU的个数
innodb_thread_concurrency = 12
innodb_max_dirty_pages_pct = 75
innodb_flush_method = O_DIRECT
innodb_purge_threads = 10
innodb_large_prefix = 1
#参数待测试
#innodb_io_capacity = 20000
#innodb_io_capacity_max = 40000
#根据CPU核心数来设定
thread_pool_size = 8
#thread_handling = pool-of-threads
thread_pool_oversubscribe = 24
#thread_handling = pool-of-threads
thread_pool_stall_limit = 100
thread_pool_max_threads = 30
#解释： 在启动时把热数据加载到内存。
innodb_buffer_pool_load_at_startup = 1
#解释： 在关闭时把热数据dump到本地磁盘
innodb_buffer_pool_dump_at_shutdown = 1
##默认是8M, 如果一次insert数据量比较多的话, 可以适当增加
innodb_autoextend_increment = 32
[mysqldump]
quick
max_allowed_packet = 512M
[mysql]
no-auto-rehash
# Remove the next comment character if you are not familiar with SQL
#safe-updates
[myisamchk]
#key_buffer_size = 20M
#sort_buffer_size = 20M
key_buffer_size = 200M
sort_buffer_size = 200M
read_buffer = 2M
write_buffer = 2M
[mysqld_safe]
#控制文件打开数的show global status like 'open%file%';比较合适的设置：Open_files / open_files_limit * 100% <= 75%
open-files-limit = 65535
log-error = /data/mysql_data/error.log
[mysqlhotcopy]
interactive-timeout
#启用存储过程、触发器的权限，但是会不适合于主从复制机制，会导致数据的不一致
log_bin_trust_function_creators = 1